Detrended fluctuation analysis and Hurst exponent as a measure to differentiate EEG signals

Abstract

This work analyze the dynamics of the electroencephalographic (EEG)
signals of normal and epileptic patients. The Detrended Fluctuation Anal-
ysis (DFA) and the Hurst exponent methods are used for estimation of the
presence of long term correlations in physiological time series observed in
healthy and unhealthy brains. The presence of long-range correlation in a
biological time series is an usually response observed in healthy organisms.
The complexity of the recorded signal can guarantee some adaptability of
the organism to the situations of disturbances. By other hand, the absence
of this correlation indicates the loss of this complexity. Non-parametric
Wilcoxon test was used for both, namely healthy and unhealthy groups, in
order to compare the mean values of the Hurst exponents. Comparison of
the means values of the Hurst exponents of normal and epileptic patients,
using the Wilcoxon test results, point to some signicant dierence between
two groups [1].